Please log in.
Before you can vote, you need to register. Please log in or create an account.
Computer: Email: Spam: Filter
Anti-spam mail server... VERY simple idea   (+1, -5)  [vote for, against]
A mail server that analyzes outgoing patterns, thus detecting/detering spam

This is a very simple idea, please poke some holes in it. :-)

There are "anti-flood" control for web forums that won't allow you to post a message within 60 seconds or so of each other. Why not put flood control on mail servers?

Have the mail server HOLD all outgoing mail for at least 15 minutes (or whatever period we choose) for each user, starting from the moment he sends his first message.

At the end of the cut-off period, check how many messages did he sent out. If it exceeds the threshold, remind him of the limit.

As an example (probably a pretty bad one), say you are limited to maximum of 15 messages per queue, and period of 15 minutes. That SHOULD be sufficient for most users. Powerusers with proven records can receive higher quotas.

If a spammer exceeds the limit, their mail server would simply refuse to accept any more outgoing mail from their account until mail server decides the queue is sufficiently clear to accept new mail.

You can even have the server run a pattern match against the mail in the queue. If they are 99% identical and/or contain common spam words, they can be dumped immediately.

This will DRASTICALLY reduce spam throughput to whatever rate the mail server sysop determines is "adequate".

If the spammers can only send one e-mail per minute, they would be driven out of business.

Sure we don't get out e-mail as fast, but wouldn't you accept the slight delay for mail nearly free of spam?

Combine this with the pay-per-message-to-send idea would be even better. The paid messages would go through immediately, but there's a per-message charge.

Comment: This wasn't meant to be a do-all end-all answer to spam, merely looking at it from a new angle. Most spam solutions concentrate on FILTERING. Why not approaching from SENDING? Thus this idea is born. I don't claim it'll work, but you must admit it LOOKS elegant. :-D

------

Additional notes:

a) snarfyguy: This type of server would only be at the local ISPs to discourage abuse by spammers who setup throwaway accounts. Of course some spam will still get through, but if you can limit the number of spam one spammer can send, you just increased his cost of doing business by several hundred (thousand) percent. By limiting their throughput, they'll be caught earlier and deactivated earlier.

It won't prevent spammers from setting up their own mail server, but that is much easier to track and cost more, which defeats the purpose of spam. Also see c)

As for paying, it's an OPTIONAL component. After all, if a spammer want to PAY to get a message to you, he must think he has something GOOD to sell, doesn't him?

b) sirrobin: nothing prevents a spammer to go to a spam-friendly ISP. However, a spam-friendly ISP will likely charge more (if they don't, they're stupid) and that again defeats the advantage of spam: almost nil operating cost.

The spam-friendly ISP can always be black-listed or filtered exclusively either through ISP level or through user-level filtering. I believe this happened to AGIS a while back (?).

c) cp: Direct-sending can be prevented by local filtering. Most "freebie" ISPs will NOT let you send SMTP mail that does NOT go through their server. I remember I was using Excite Freelane and I absolutely couldn't send mail through Outlook Express (Excite suggests that I use webmail instead). I finally found their ISP and their instructions on how to use THEIR SMTP server to get mail out. If you can FORCE all SMTP traffic to go through YOUR SMTP server, you have control over all that traffic, and it also provides more tracking.

If you want to setup your own mail server, you can always petition the ISP to setup a special case for you, and your activity will be logged and perhaps auditted periodically.

d) Rods Tiger: One mail with hundreds/thousands of recipients is too easy to filter. On USENET you can filter out posts that were crossposted to too many groups. You can do the same in e-mail. I believe some e-mail server filters this as well depending on local config. Most spammers limit themselves to like 10-20 recipients per e-mail, which means they have to send out hundreds if not thousands of e-mail to reach their "audience". By slowing down their throughput their spam will be caught earlier, as complaints are received sooner and processed sooner, and they got deactivated sooner, thus increasing their cost of doing business.

Good comments, please continue. :-)
-- kschang, Jan 20 2002

I'm all about getting rid of spam, but how would spam be differentiated from legitimate mass-mailings in this system? I, for example, give a party every month or so, notifying some invitees by e-mail. These recipients have all signed a mailing list in order to receive such notifications, about 500 addresses in total (so it's *literally* opt-in). What would happen to me?

Paying to send e-mail doesn't appeal to me, or anyone else, I'd wager...
-- snarfyguy, Jan 20 2002


What stops a spammer from taking their business to a provider that doesn't stop spammers?
-- sirrobin, Jan 20 2002


Okay, I have some holes for you:

The problem with this idea is that you don't need your ISP's mail server to send mail. It is trivial to look up an MX record and send a message manually, so this mechanism would only stop a certain person sending a bunch of messages to a certain domain (e.g. hotmail.com).

I have a friend who runs his own mail server for both incoming and outgoing mail as he doesn't trust his ISP; if a pay-per-send system was attempted, I would do the same.
-- cp, Jan 20 2002


Depends. Sometimes not; it can be hundreds of email with 20 or so addresses, instead of one mail with 2000 addresses, to try and get around limitations. I get spam with 20 addresses, all beginning with 'star' a lot.
-- StarChaser, Jan 21 2002



random, halfbakery